Wenkai Yang

This article examines the threat of backdoor attacks on language model agents.

Research reveals significant security risks in chat models from backdoor attacks.

Explores the challenges of supervising advanced AI models with weaker counterparts.